Duke Machine Learning

Welcome to the home page for the Duke STA561/COMPSCI571 Kaggle competition! We’ll be working with a subset of data from an existing Kaggle contest (https://www.kaggle.com/c/allstate-claims-severity). We recommend checking out some of the forum posts and discussions for that Kaggle contest as they will likely be very relevant to this one as well. Your task is to accurately predict the severity of insurance claims as measured in US dollars. Note that while there is data overlap with the main Kaggle competition ongoing, the data ID and rows have been randomized so it won’t be possible to gain an advantage by snooping on the larger, publicly available dataset. Your submissions will be CSV files with one column labeled “id” and another column labeled “loss”. The ID will correspond to the row of…


Link to Full Article: Duke Machine Learning